Algorithms and software for multilevel nonlinear optimization

  • Dimitri Tomanos

    Student thesis: Doc typesDoctor of Sciences

    Abstract

    This research concerns the algorithmic study of multilevel nonlinear optimization problems. We have developped during our thesis new methods to solve this type of problems and we analyze their convergence and numerical results. We have also implemented a soft- ware for which a complete description is given in this manuscript. A library of test problems has also been constructed on which we have tested our software. The numerical results ob- tained show that our method is clearly the best for this class of problems. The software also allowed the solution of a progressive lens industrial application that had never been solved by existing methods.
    Date of Award4 Sept 2009
    Original languageEnglish
    Awarding Institution
    • University of Namur
    SupervisorPhilippe Toint (Supervisor), Annick Sartenaer (President), Joseph Winkin (Jury), Serge Gratton (Jury) & Dominique Orban (Jury)

    Cite this

    '